Hi.
I am creating very short (1-5 seconds) mp3 clips for a language-learning app, the same type of thing as DuoLingo.
Could anyone advise on best practice? The results so far are decent but not uniform, and sometimes quite sub-par.
My process:
Noise Reduction: about 15 dB - sounds good at 50% volume, but gets noisy at 75% volume - should I use a higher dB ? I'm scared to go too high, as I know the voice can start to sound tinny and metallic artefacts can appear.
Amplification: about 20 db, to bring amplify the wave to around +7 / - 0.7 dB ie: 3/4s height - is this good practice?
Lead-in and tail-off : I leave 1/4 of a second - I'm reluctant to lengthen the lead-in especially, as the clip needs to play without ant delay (again, think DuoLingo) - but is 1/4 second too tight?
Fade-in, fade-out : usually do a double fade-in and a single fade-out. I find this helps, as any recording noises are most likely to be at the start or end
Voice clicks : a scourge - occur on certain sounds, and in some clips, I cannot find a way to remove them. How do DuoLingo manage to remove theirs ??
Is there a way, using Noise Reduction, to get the profile of a click, and thus remove them all from a clip?
Would really appreciate advice from someone who's been down this road, and can help me produce top quality clips for my forthcoming French and English courses.
Merci.
Best practice for creating 1-5 second mp3 clips
Forum rules
This forum is for Audacity on Windows.
Please state which version of Windows you are using,
and the exact three-section version number of Audacity from "Help menu > About Audacity".
Audacity 1.2.x and 1.3.x are obsolete and no longer supported. If you still have those versions, please upgrade at https://www.audacityteam.org/download/.
The old forums for those versions are now closed, but you can still read the archives of the 1.2.x and 1.3.x forums.
Please state which version of Windows you are using,
and the exact three-section version number of Audacity from "Help menu > About Audacity".
Audacity 1.2.x and 1.3.x are obsolete and no longer supported. If you still have those versions, please upgrade at https://www.audacityteam.org/download/.
The old forums for those versions are now closed, but you can still read the archives of the 1.2.x and 1.3.x forums.
Re: Best practice for creating 1-5 second mp3 clips
For consistency you may be best to make one long recording of multiple phrases, process the recording as one, then split the recording into individual MP3s (see: https://manual.audacityteam.org/man/spl ... racks.html)
9/10 questions are answered in the FREQUENTLY ASKED QUESTIONS (FAQ)
Re: Best practice for creating 1-5 second mp3 clips
Every situation is different and noise reduction is just one of those things you have to experiment with. Ideally, you wouldn't need it and if the noise is bad "the cure can be worse than the disease."My process:
Noise Reduction: about 15 dB
As long as you don't go over 0dB boosting the volume digitally doesn't change the quality or signal-to-noise ratio. It's the same as the listener turning-up the analog volume control. Of course when the volume is turned-up the background noise gets turned-up along with the signal so it seems worse.- sounds good at 50% volume, but gets noisy at 75% volume - should I use a higher dB ? I'm scared to go too high, as I know the voice can start to sound tinny and metallic artefacts can appear.
That's an indication that your recording levels are low and there might be something "wrong" with your set-up... Maybe you need to get closer to the mic or speak louder or something... Or, maybe you just need to turn-up the recording volume. Digital levels aren't that critical (as long as you don't "try" to go over 0dB and clip/distort) but peaks of -20dB are on the low side. Typically you should shoot-for around -6dB (peak).Amplification: about 20 db,
Your "numbers" are confusing... 75% is about -3dB, 50% is -6dB, and 10% is -20dB. It's OK to accept the Amplify default and hit 0dB on the peaks. But some clips will sound louder than others so you may need to bring-down the louder-sounding clips.*to bring amplify the wave to around +7 / - 0.7 dB ie: 3/4s height - is this good practice?
That's an "artistic decision".Lead-in and tail-off : I leave 1/4 of a second - I'm reluctant to lengthen the lead-in especially, as the clip needs to play without ant delay (again, think DuoLingo) - but is 1/4 second too tight?
Fade-in, fade-out : usually do a double fade-in and a single fade-out. I find this helps, as any recording noises are most likely to be at the start or end
No. Regular noise reduction is for constant background noise. There is a (automatic) Click Removal effect and a (manual) Repair effect but it might be easier just to re-record those clips.Voice clicks : a scourge - occur on certain sounds, and in some clips, I cannot find a way to remove them. How do DuoLingo manage to remove theirs ??
Is there a way, using Noise Reduction, to get the profile of a click, and thus remove them all from a clip?
---------------------------------
You might take a look at the Recommended Audiobook Mastering Process. Of course you don't have to exactly meet those standards but all of your clips would be the same loudness and if you can hit the noise spec you know the noise is good enough for professional production.
Since you're not submitting and audiobook to ACX and I suggest a couple of changes - Set the limiter to 0dB, set the RMS level ("loudness") correspondingly 3.5dB louder to -16.5dB. Of course, that 3.5dB boost will make the ACX noise floor measurement 3.5dB "worse" but it's not really worse because you haven't changed the signal-to-noise ratio.
-
kozikowski
- Forum Staff
- Posts: 69374
- Joined: Thu Aug 02, 2007 5:57 pm
- Operating System: macOS 10.13 High Sierra
Re: Best practice for creating 1-5 second mp3 clips
You're going to experience all the audiobook problems. "I can hear traffic sounds, my background noise is too loud, I have clicky sounds, I sound like I'm recording in a bathroom," etc.
You in particular probably shouldn't be recording anything on the computer.
This is an affordable home studio with a stand-alone sound recorder. Zoom H4 pictured. I have used a Zoom H1n as well.
viewtopic.php?p=369938#p369938
The audiobook mastering tools solve a lot of problems. The goal is to crank out chapter after chapter with everything the same volume. That sounds a lot like what you're doing.
https://wiki.audacityteam.org/wiki/Audiobook_Mastering
This is the short form.

A note here that MP3 files don't have precise, standard stop and start times. For example, people trying to produce loops, a repeating sound track where the beginning and ending match almost always fail when they try to do it in MP3. You don't have the precision requirement, but still. That's going to happen.
There is another MP3 problem as well. The sound quality gets worse as you edit and do production. Do everything in WAV (Microsoft) 16-bit until you get to producing the actual product clips. Then make the MP3s.
And yes, as Steve above, record the work in large blocks instead of trying to do it phrase by phrase. That will increase the chances of a smooth flow and allow you to blanket apply the mastering tools.
Koz
You in particular probably shouldn't be recording anything on the computer.
This is an affordable home studio with a stand-alone sound recorder. Zoom H4 pictured. I have used a Zoom H1n as well.
viewtopic.php?p=369938#p369938
The audiobook mastering tools solve a lot of problems. The goal is to crank out chapter after chapter with everything the same volume. That sounds a lot like what you're doing.
https://wiki.audacityteam.org/wiki/Audiobook_Mastering
This is the short form.

A note here that MP3 files don't have precise, standard stop and start times. For example, people trying to produce loops, a repeating sound track where the beginning and ending match almost always fail when they try to do it in MP3. You don't have the precision requirement, but still. That's going to happen.
There is another MP3 problem as well. The sound quality gets worse as you edit and do production. Do everything in WAV (Microsoft) 16-bit until you get to producing the actual product clips. Then make the MP3s.
And yes, as Steve above, record the work in large blocks instead of trying to do it phrase by phrase. That will increase the chances of a smooth flow and allow you to blanket apply the mastering tools.
Koz
Re: Best practice for creating 1-5 second mp3 clips
I've studied all your replies. Lots of good stuff in there. Thanks for your input.